Pedagogical Use of Generative Artificial Intelligence
Generative artificial intelligence is developing rapidly, providing opportunities and challenges for teaching and learning. IDEA aims to promote a curious, creative, and responsible use of GenAI tools.
Ambitions for the Use of Generative Artificial Intelligence at UV
Generative artificial intelligence is a specific subcategory within artificial intelligence that focuses on generating new data or content based on existing data. This can include text generation, image creation, music, and other forms of creative content. Many students already use various AI tools in their academic work today, and it is crucial that students develop competence in responsible AI use as part of their studies.
Holistic Learning Design
In a holistic learning design, the use of generative artificial intelligence should be included so that students have the opportunity to explore the technology as part of their learning work. This highlights assessment practices. When we talk about assessment, it is important to distinguish between ongoing assessment (=formative) and final assessment (=summative). Both are part of a holistic learning design.
Self-reflection in Teaching
Students' use of generative artificial intelligence in their learning work should be followed up with discussions related to the potential and challenges posed by the technology. This enables students and teachers to have common discussions about the fair and ethical use of technology. Formative assessment forms are well-suited to facilitate students' self-reflection.
Final Assessment
To prevent uncritical and dishonest use of generative AI tools in final assessments (=assignments and exams), these should include elements such as reflection notes or other forms of self-evaluations that contribute to reflection among students regarding their own learning process.
All summative assessments follow established guidelines and frameworks for exam answers and written assignments at UiO. What may lead to suspicion of cheating/attempted cheating is described in Regulations on Studies and Examinations at the University of Oslo (§ 7.6).
Since UV encourages the use of generative AI tools also in final assessments, it must be made clear to students how they should document their use in assignment and exam answers. This is done at the course level and must be described on the course page. Additionally, it must be stated both in the assessment criteria and in the examiner's guidance how students' potential use of generative artificial intelligence in summative assessments will be evaluated. This responsibility lies with the course responsible.
Self-declaration of AI use in Final Assessment
Students can submit a self-declaration, and UV has developed a specific form that can be adapted to each course. The form can either be uploaded along with an assignment in Inspera or included as a declaration in Inspera where students specify their AI use. It is the responsibility of the course responsible to assess permitted AI use in the course and to adapt the declaration for use in the course.
You can find the self-declaration form here (Norwegian only)
What Are the Differences Between UiO-Approved GenAI Tools?
UiO offers four different AI tools that are approved for academic use: GPT UiO, Microsoft Copilot, Notebook LM and Autotekst. Below are the key features and data approval levels for each tool.
GPT UiO
GPT UiO is UiO's own privacy-friendly large language model (LLM) service. It generates responses based on the content it receives. This is often referred to as a "prompt" or an instruction.
- Approved for green and yellow data.
Microsoft Copilot
Microsoft Copilot is another LLM (large language model) that works similarly to GPT UiO, responding to your requests by generating a response. Copilot can search the web in real-time and provides links to the sources of information.
- Approved for use of green data.
You can find Microsoft Copilot here
Gemini NotebookLM (formerly NotebookLM)
NotebookLM is an AI tool that can create summaries, explanations, answers and more using user-uploaded content: text files, PDFs, websites, YouTube videos, documents and slides. It can also generate “audio summaries,” which summarize information in a conversation-based “podcast” using AI-generated voices.
- Approved for green and yellow data.
- For privacy and data security reasons, you must use UiO login through Google Workspace. This way, your personal data is not used to further develop the model, and your data is stored in European data centres. You can read more about how to get this here.
- The service requires you to have an account in Google Workspace through UiO. You can read more about how to obtain this here.
- UiO does not have its own guide for using NotebookLM, but Google has a dedicated guide that is helpful.
Autotekst
Autotekst is a speech-to-text transcription service that converts audio or video files into text format. It supports multiple languages and can also translate audio content into English text files.
- Autotekst is approved for green, yellow, and red data.